S41467-018-05074-Y.Pdf
Total Page:16
File Type:pdf, Size:1020Kb
ARTICLE DOI: 10.1038/s41467-018-05074-y OPEN Identification of susceptibility pathways for the role of chromosome 15q25.1 in modifying lung cancer risk Xuemei Ji et al.# Genome-wide association studies (GWAS) identified the chromosome 15q25.1 locus as a leading susceptibility region for lung cancer. However, the pathogenic pathways, through which sus- 1234567890():,; ceptibility SNPs within chromosome 15q25.1 affects lung cancer risk, have not been explored. We analyzed three cohorts with GWAS data consisting 42,901 individuals and lung expression quantitative trait loci (eQTL) data on 409 individuals to identify and validate the underlying pathways and to investigate the combined effect of genes from the identified susceptibility pathways. The KEGG neuroactive ligand receptor interaction pathway, two Reactome pathways, and 22 Gene Ontology terms were identified and replicated to be significantly associated with lung cancer risk, with P values less than 0.05 and FDR less than 0.1. Functional annotation of eQTL analysis results showed that the neuroactive ligand receptor interaction pathway and gated channel activity were involved in lung cancer risk. These pathways provide important insights for the etiology of lung cancer. Correspondence and requests for materials should be addressed to C.I.A. (email: [email protected]). #A full list of authors and their affliations appears at the end of the paper. NATURE COMMUNICATIONS | (2018) 9:3221 | DOI: 10.1038/s41467-018-05074-y | www.nature.com/naturecommunications 1 ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-05074-y ung cancer, accounting for 13% of all cancer cases and 23% Results Lof all cancer-related deaths worldwide, is a leading cause of The study design is presented in Fig. 1. Demographic char- cancer death in the US and around the world, and represents acteristics and sample sizes of the two discovery cohorts and the a major public health problem1. Several genome-wide association replication cohort for GWAS pathway analyses are summarized studies (GWAS) have been published and identified the chro- in Table 1. Demographic characteristics of the lung eQTL study mosome 15q25.1 locus as a susceptibility region for lung cancer2– are summarized in Supplementary Table 1. 4, smoking behavior5,6, and nicotine addiction4 in Caucasians2, African-Americans7, and Asians8. Epigenetic analyses provided Selection of index SNPs and candidate SNPs in discovery.To evidence that epigenetic silencing of nAChR-encoding genes determine the most important susceptibility loci for lung cancer clustered at the 15q25.1 locus may contribute to lung cancer risk9. and to identify multiple association signals within observed loci, In addition, expression quantitative trait loci (eQTL) studies we performed association analyses using the 1st and 2nd dis- showed an influence of alleles in this region on the expression of covery cohort, separately, and conducted a meta-analysis of the several genes at chromosome 15q25.1, providing a mechanism by two cohorts in the discovery phase. We identified the most sig- which these variations might affect lung cancer risk10. Our pre- nificant susceptibility loci for lung cancer on chromosomes vious GWA studies found that variants in chromosome 15q25.1, 15q25.1 in both discovery cohorts, and confirmed the finding in including single-nucleotide polymorphisms (SNPs) and haplo- the meta-analysis. Eight signals within chromosome 15q25.1 were types, are involved in the etiology of overall lung cancer sus- defined as lung cancer risk-associated SNPs based on P values of ceptibility and by histology and smoking status2,11. However, − association with lung cancer of less than 5 × 10 8 in the 1st lung cancer, being a disease of complex origin, is usually con- discovery cohort and in the meta-analysis (Table 2). After Bon- sidered to result from complex effects of smoking along with ferroni correction, the eight signals maintained a significant multiple genetic variants affecting a number of pathways or impact on lung cancer risk in the 1st discovery cohort and in the biological process. Common SNPs are not individually known to meta-analysis. We defined the eight significant SNPs, which were add greatly to individual risk, unless more complex gene–gene rs1051730 in CHRNA3; rs1996371, rs6495314, rs11638372, interactions play a crucial role in the genetic architecture of rs4887077, and rs6495309 in CHRNB4; and rs8034191 and pathogenesis of complex disorders12, such as lung cancer. The rs2036534 in HYKK, as the index SNPs for lung cancer risk, and pathogenic pathways, through which lung cancer susceptibility used these eight SNPs to further select the candidate SNPs, which SNPs within chromosome 15q25.1 affect disease etiology and interacted with the eight index SNPs. development of lung cancer, have not been studied comprehen- To evaluate potential functional connections between genes sively, limiting mechanistic understanding. mapping throughout the genome and those on chromosome The objective of this study was to explore the underlying 15q25.1, to further elucidate the role of the chromosome 15q25.1- pathways that are involved in the molecular mechanisms by related pathway in lung cancer risk, we investigated SNP–SNP which variants at the chromosome 15q25.1 locus modify lung interactions between the eight index SNPs within chromosome cancer risk and increase lung cancer occurrence and develop- 15q25.1 and the whole genome in both discovery cohorts, and ment. We first performed a GWAS analysis with a cohort of conducted a meta-analysis of SNP–SNP interaction with both 1923 lung cancer cases and 1977 healthy controls of Italian cohorts. A total of 5883 SNP pairs between the eight index SNPs origin combined with a cohort of 2995 lung cases and 3578 and candidate SNPs in the whole genome exhibited epistasis P value controls of European ancestry, and then conducted a meta- of less than 0.05 in the 1st discovery cohort and in the meta-analysis analysis to identify the index SNPs within the chromosome results and showed epistasis P value of less than 0.10 in the 2nd 15q25.1 locus that were significantly associated with lung discovery cohort (Supplementary Data 1). In total, 3409 candidate cancer risk. We then investigated the SNP–SNP interaction SNPswithinthewholegenomewereidentified and validated to between the index SNPs within the chromosome 15q25.1 locus interact with the eight index SNPs (Supplementary Data 2). and the entire genome to identify the SNPs that interact with the 15q25.1 index SNPs, and are therefore involved in lung cancer etiology through interaction. Furthermore, using the Susceptibility pathways and GO terms in discovery. In order to index SNPs and their related SNPs in the whole genome, we identify chromosome 15q25.1-associated pathogenic pathways explored the pathogenic pathways that may be relevant to lung and biological processes that may be relevant to lung cancer cancer etiology, and replicated the findings with an indepen- etiology, we then conducted enrichment analyses using i- dent cohort of 18,439 lung cancer cases and 14,026 healthy GSEA4GWAS13 in discovery phase with the meta-analysis controls. We also studied genome-wide gene expression data in results of 2530 SNPs, which were pruned from eight index human lung tissues and conducted an eQTL analysis to inves- SNPs and 3409 candidate SNPs for linkage disequilibrium (LD) tigate whether the functional annotation of the eQTL results to reduce the possibility of biased results. We applied mapping can validate the susceptibility pathways from our GWAS ana- rules of SNPs to genes by incorporating a region 20 kb upstream lyses. Finally, we explored whether genes from our suscept- and downstream of each gene (Supplementary Data 2 and 3). In ibility pathways might jointly affect the process by which the total, one Kyoto Encyclopedia of Genes and Genomes (KEGG) chromosome 15q25.1 locus influences lung cancer risk. Our pathway, three Reactome pathways, and 22 Gene Ontology (GO) findings suggest that common genetic variations within chro- terms were significantly associated with lung cancer risk with mosome 15q25.1 are likely to affect lung cancer etiology by improved gene set enrichment analysis (i-GSEA)13 P values less influencing the expression/structure and thereby the function of than 0.05 and FDR less than 0.25 for each pathway (Table 3). The genes that comprise the neuroactive ligand receptor interaction KEGG pathway was the neuroactive ligand receptor interaction pathway or gated channel activity and related terms. Such new pathway (i-GSEA P = 0.001 and FDR = 0.006). The 22 GO terms biologic insights from pathway analysis will provide a better included substrate-specific channel activity (i-GSEA P < 0.001 understanding of the etiology and development of lung cancer, and FDR = 0.005), ion channel activity (i-GSEA P < 0.001 and potentially shortening the interval between increasing biologic FDR = 0.005), gated channel activity (i-GSEA P = 0.002 and knowledge and translation to patient care. FDR = 0.006), and several similar terms. 2 NATURE COMMUNICATIONS | (2018) 9:3221 | DOI: 10.1038/s41467-018-05074-y | www.nature.com/naturecommunications NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-05074-y ARTICLE Discovery 1st cohort : 1977 cases and 1923 controls with 561, 466 SNPs 1 Primary GWAS analysis Discovery 2nd cohort : 2995 cases and 3578 controls with 310, 276 SNPs Meta–analysis of the two discovery cohorts with 310, 276 SNPs Chromosome 15q25. 1 SNPs Association P value < 5 × 10–8 in the discovery 1st cohort 2 Index